Assembly and Preferential Localization of Nup116p on the Cytoplasmic Face of the Nuclear Pore Complex by Interaction with Nup82p

Abstract

The yeast Saccharomyces cerevisiae nucleoporin Nup116p serves as a docking site for both nuclear import and export factors. However, the mechanism for assembling Nup116p into the nuclear pore complex (NPC) has not been resolved. By conducting a two-hybrid screen with the carboxy (C)-terminal Nup116p region as bait, we identified Nup82p. The predicted coiled-coil region of Nup82p was not required for Nup116p interaction, making the binding requirements distinct from those for the Nsp1p-Nup82p-Nup159p subcomplex (N. Belgareh, C. Snay-Hodge, F. Pasteau, S. Dagher, C. N. Cole, and V. Doye, Mol. Biol. Cell 9:3475–3492, 1998). Immunoprecipitation experiments using yeast cell lysates resulted in the coisolation of a Nup116p-Nup82p subcomplex. Although the absence of Nup116p had no effect on the NPC localization of Nup82p, overexpression of C-terminal Nup116p in a nup116 null mutant resulted in Nup82p mislocalization. Moreover, NPC localization of Nup116p was specifically diminished in a nup82-Δ108 mutant after growth at 37°C. Immunoelectron microscopy analysis showed Nup116p was localized on both the cytoplasmic and nuclear NPC faces. Its distribution was asymmetric with the majority at the cytoplasmic face. Taken together, these results suggest that Nup82p and Nup116p interact at the cytoplasmic NPC face, with nucleoplasmic Nup116p localization utilizing novel binding partners.

Nuclear pore complexes (NPCs) are massive multiprotein structures embedded in the nuclear envelope (NE), which serve as portals for regulating the traffic of macromolecules between the cytoplasm and the nucleus (52). Three-dimensional structural information for yeast Saccharomyces cerevisiae and vertebrate NPCs has been recently revealed by a combination of high-resolution cryoelectron and scanning electron microscopy (EM) analysis (2, 3, 16, 27, 46, 47, 69). NPCs possess a central plug surrounded by eight spokes which attach to cytoplasmic and nuclear rings. These rings anchor the peripherally associated cytoplasm filaments and nuclear basket. Yeast S. cerevisiae NPCs are comprised of ∼30 different proteins, termed nucleoporins (48). Models of NPC structure have predicted that the distinct modular structures of the NPC are formed from subsets of distinct nucleoporin subunits. In support of this, subcomplexes containing different nucleoporins have been biochemically isolated and characterized (for example, references 6, 9, 14, 21 to 23, 25, 30, 40, 41, and 58). Moreover, immunoelectron microscopy (IEM) experiments have localized some nucleoporins to exclusively the cytoplasmic filaments or the nuclear basket and others to either symmetric or asymmetric distributions on the central core structure (reviewed in references 48 and 60). This differential localization may reflect distinct roles for particular nucleoporins in mediating particular steps of the nuclear transport mechanism.

One strategy for dissecting the hierarchy of protein-protein interactions that account for NPC structure and function has been to analyze yeast S. cerevisiae mutants. Mutations in a number of yeast genes encoding nucleoporins result in perturbations of nuclear transport and of NE-NPC structure (for reviews, see references 10 and 65). Structural perturbations include clustering of NPCs to localized patches of the NE, decreased NPC number per nucleus, the formation of NE herniations over the cytoplasmic face of the NPC, NE projections, blisters of the NE, the presence of intranuclear annulate lamellae, and extensive lobulation of the NE. Mutants with altered NPC stoichiometry for distinct nucleoporins have also been characterized (6, 8, 37). Overall, work to date supports a structural framework wherein a network of in vivo interactions mediate NPC biogenesis and structural integrity.

A subset of nucleoporins are peripherally positioned to interact with shuttling nuclear transport factors (reviewed in references 42, 52, and 68). Our previous studies have focused on characterization of the nucleoporin Nup116p, a member of the GLFG family of nucleoporins, which harbors a region containing repeats of the tetrapeptide glycine-leucine-phenylalanine-glycine (GLFG) separated by essentially uncharged spacer sequences enriched in glutamine, asparagine, serine, and threonine (66, 67). Although Nup116p is not essential for cell viability (66), nup116 null (nup116Δ) mutants are temperature sensitive for growth (64). The lethal phenotype type correlates with defects in mRNA export and perturbations of NPC-NE structure (64). Molecular dissection of the structural regions of Nup116p has identified at least three functional domains, each of which is required for normal growth (33). The GLFG region directly interacts with a member of the karyopherin/importin/exportin/transportin family of nuclear transport factors (33, 34). The amino (N)-terminal domain of Nup116p contains FG repeats and serves as a docking site for the shuttling mRNA export factor Gle2p (4, 28, 44). Thus, Nup116p interacts with both import and export factors and plays a key role in nuclear transport.

Understanding how Nup116p is assembled into the NPC is important for further resolving its role in mediating the movement of import and export factors or substrates through the portal. A previous report has shown that the carboxy (C)-terminal region of Nup116p (Nup116-C) can be independently targeted to the NPC (4). The mechanism for such assembly by Nup116-C has not been elucidated. Moreover, protein-protein interaction partners for Nup116-C have not been defined. This region is homologous to regions in two other GLFG nucleoporins: the middle (M) region of Nup145p (Nup145-M) and the C-terminal region of Nup100p (Nup100-C) (12, 63, 66). Each of these regions harbors a peptide octamer that others have suggested is necessary for in vitro binding to homopolymeric RNA of guanine residues [poly(G) (12)]. The amino acid octamer has been designated the nucleoporin RNA-binding motif (NRM). In addition to a potential RNA-binding function, the Nup116-C, Nup100-C, and Nup145-M regions also mediate a gain-of-function lethal phenotype (28). When these regions are expressed in the absence of full-length Nup116p, yeast cells are not viable.

We have conducted a two-hybrid screen with Nup116p-C to identify protein interaction partners. A specific interaction with the essential nucleoporin Nup82p was identified, and further experiments demonstrated that Nup116-C associates in vivo and in vitro with Nup82p. Immunofluorescence experiments suggested that Nup82p serves to anchor Nup116p at the cytoplasmic face of the NPC, and IEM analysis showed Nup116p at both faces of the NPC, with the majority at the cytoplasmic face. These results yield important mechanistic insights into how Nup116p may facilitate nuclear import and export.

MATERIALS AND METHODS

Strains and plasmids.

The plasmids used in this study are described in Table 1. Bacterial strains were cultured in SOB medium and transformed by standard methods (53). Escherichia coli strain DH5α was used as the bacterial host for all plasmids. The yeast strains were grown in either rich medium (YPD; 1% yeast extract, 2% Bacto Peptone, 2% glucose) or synthetic minimal (SM) medium supplemented with appropriate amino acids and 2% glucose. Yeast transformations were performed by the lithium acetate method (35), and general yeast manipulations were conducted as described elsewhere (57). The haploid yeast strains used in this study were W303α (MATα ade2-1 ura3-1 his3-11,15 trp1-1 leu2-3,112 can1-100), SWY27 (nup116Δ [64]), SWY1441 (NUP82-GFP [8]), SWY1695 (GFP-NIC96 [8]), SWY1976 (nup116Δ NUP82-GFP), SWY2126 (nup82-Δ108 GFP-NIC96), a nup82-Δ108 strain (31), and PJ69-4A (36).

TABLE 1.

Plasmids used in this study

Plasmida	Construction	Reference
pGAD backbone
pSW1126	Full-length NUP82 locus in BamHI site	This study
pSW1212	Fragment encoding amino acids 1–187 of NUP82 locus in BamHI site	This study
pSW1213	Fragment encoding amino acids 1–254 of NUP82 locus in BamHI site	This study
pSW1215	Fragment encoding amino acids 1–409 of NUP82 locus in BamHI-PstI	This study
pSW1217	Fragment encoding amino acids 1–651 of NUP82 locus in BamHI-PstI	This study
pSW1218	Fragment encoding amino acids 351–713 of NUP82 locus in ClaI-BglII	This study
pSW1219	Fragment encoding amino acids 482–713 of NUP82 locus in ClaI-BglII	This study
pSW1216	Full-length NSP1 locus in BamHI-PstI	This study
pGBT8 (GBD) backbone
pSW259	Fragment encoding amino acids 2–664 of NUP145 locus in BamHI-SacI	11
pSW413	Full-length GLE1 locus in BamHI	This study
pSW423	Fragment encoding amino acids 590–959 of NUP100 locus in BamHI-SacI	This study
pSW546	Fragment encoding amino acids 726–1113 of NUP116 locus in BamHI-SacI	This study
pSW547	Fragment encoding amino acids 726–919 of NUP116 locus in BamHI-SacI	This study
pSW548	Fragment encoding amino acids 914–1113 of NUP116 locus in BamHI-SacI	This study
pSW560	Fragment encoding amino acids 914–1113 of NUP116 locus with 998–1105 replaced by G in BamHI-SacI	This study
pGBD backbone
pSW1211	Fragment encoding amino acids 998–1105 of NUP116 locus in BamHI-SalI	This study
pSW1222	Full-length NUP159 locus in EcoRI	This study
pBJ382 backbone
pSW171	Fragment encoding amino acids 726–1113 of NUP116 locus in BamHI-SacI	28

^aVector backbone references (in parentheses): pG_AD and pG_BD (41); pGBT (5); pBJ382 (gift from C. Hug, Washington University, St. Louis, MO); pCITE (Novagen). 

Two-hybrid screen.

The two-hybrid host strain PJ69-4A harboring pSW546 (coding sequence for Nup116p-C fused to the Gal4p DNA-binding domain [G_BD–Nup116-C; residues 726 to 1113]) was transformed with three pools of a yeast genomic library that contains sequences fused to the Gal4p activation domain (pG_AD-C1, -C2, and -C3; provided by P. James) (36). Approximately 2.2 × 10⁶ transformants of pool pG_AD-C1, 2.1 × 10⁶ of pG_AD-C2, and 1.8 × 10⁶ of pG_AD-C3 were screened on medium lacking histidine. Positively interacting colonies were retested on medium lacking adenine and assayed for expression of β-galactosidase (7). Specificity of the interaction was tested by expressing positive library clones with lamin C fused to G_BD (Stratagene, La Jolla, Calif.). We identified one positive clone in library pool pG_AD-C1, seven in pool pG_AD-C2, and six in pool pG_AD-C3. NUP82 was present in four of the positive clones from the pG_AD-C2 pool. The other genes isolated were CIN5 (twice), RDN37 (twice), HAL5 (once), ALG7 (once), DAL81 (once), unknown open reading frame (once), YJL145W, telomere sequence (once), and sequence between PTR3 and MET10 (once).

Whole cell lysates, metabolic labeling, and immunoprecipitation.

Cultures (50 ml) of wild-type, NUP82-GFP (SWY1441), or GFP-NIC96 (SWY1695) cells were grown to an optical density at 600 nm (OD₆₀₀) of 0.5 in YPD or SM medium lacking methionine. For metabolic labeling, 100 μCi of [³⁵S]methionine (ICN) was added, and growth continued for 1 h. Cells were harvested, washed in H₂O, and resuspended in 300 μl of ice-cold lysis buffer supplemented with Complete protease inhibitor cocktail (Boehringer Mannheim, Indianapolis, Ind.); 500 μl of glass beads was added, followed by vortexing (four 1-min pulses, 2-min rests). An additional 350 μl of lysis buffer was then added, with vortexing for another minute. The total cell extract was isolated after centrifugation for 10 min at 3,000 rpm at 4°C. For immunoprecipitations, 100 μl of cell extract was mixed with the appropriate antibody (4 μl of anti-GLFG antibody, 8 μl of affinity-purified rabbit polyclonal antibody raised against Nup116-C [WU600 {33}], or 8 μl of preimmune serum) and 50 μl of packed protein A-Sepharose beads. After incubation for 90 min at 4°C, the beads were isolated by centrifugation and washed six times with 0.5 ml of ice-cold wash buffer. Immunoprecipitates were eluted in 25 μl of sodium dodecyl sulfate (SDS) sample buffer and boiled. Unbound fractions were methanol precipitated and resuspended in SDS sample buffer.

Protein samples were analyzed by polyacrylamide gel electrophoresis (PAGE) in SDS–7% polyacrylamide gels, followed by transfer to nitrocellulose membranes. Blots were probed with mouse anti-green fluorescent protein (GFP) monoclonal antibody (MAb) (Clontech, Palo Alto, Calif.) at a 1:500 dilution (16 h, 4°C) or anti-Nup116-C antibody at a 1:2,500 dilution (1 h, 23°C). All dilutions were made in 10 mM Tris-HCl (pH 8.0)–150 mM NaCl–0.05% Tween 20 (TBST)–2% nonfat dry milk. After washing in TBST, blots were incubated with peroxidase-labeled anti-mouse immunoglobulin G (IgG) or anti-rabbit IgG (diluted 1:2,000; Amersham, Arlington Heights, Ill.) for 1 h. Blots were developed by enhanced chemiluminescence (Amersham Pharmacia Biotech, Piscataway, N.J.).

Fluorescence and immunofluorescence microscopy.

Indirect immunofluorescence experiments were performed as described elsewhere (66). Wild-type or nup82-Δ108 yeast cells in early log phase were grown at 23°C or shifted to 37°C for 3.5 h and fixed for 10 min in 3.7% formaldehyde–10% methanol. Samples were incubated with mouse MAbs generated against Pom152p (MAb 118C3 [61]) or Nup159p (MAb 165C10 [39]) for 16 h at 4°C. This was followed by incubation with anti-Nup116-C rabbit antibodies (1:2,500) for 1 h at room temperature. Samples were washed with M buffer (40 mM K₂HPO₄, 10 mM KH₂PO₄, 150 mM NaCl, 0.1% NaN₃, 0.1% Tween 20, 2% nonfat dry milk). Bound antibodies were detected with affinity-purified fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse and Texas red-conjugated goat anti-rabbit antibodies (Cappel Laboratories, Organon Teknika Corp., Durham, N.C.) (1:200, 1 h, 23°C). After additional washes in M buffer and 1% bovine serum albumin (BSA)–phosphate-buffered saline (PBS), cells were mounted in 90% glycerol–1 mg of p-phenylenediamine per ml (pH 8.0) with 0.05 μg of 4′,6-diamidino-2-phenylindole (DAPI) per ml. Wild-type, nup116Δ, nup116Δ/GAL-NUP116-C, and nup116Δ/GAL cells expressing Nup82-GFP or wild-type and nup82-Δ108 cells expressing GFP-Nic96 were grown to early log phase in SM medium at 23°C or shifted to 37°C for 3.5 h and examined by direct fluorescence microscopy. Images were collected with a charge-coupled device digital camera (Dage MTI, Michigan City, Ind.).

Immunoblot analysis of Nup116p in nup82-Δ108 cells.

Mutant nup82-Δ108 cells were grown to an OD₆₀₀ of ∼0.5 and then shifted to 37°C for the indicated time. The nup116Δ/GAL-NUP116-C and nup116Δ/GAL-control cells were grown to an OD₆₀₀ of ∼0.3 in glucose medium, and cells were shifted to galactose medium for 3 h. Total yeast cell lysates were prepared from 25 mg of cells, and samples were separated by PAGE on SDS–7% polyacrylamide gels. The samples were transferred to nitrocellulose and probed with either anti-Nup116-C antibody (1:500), anti-GLFG antibody (1:2,000), or mouse MAb 12CA5 (1:1,000 for 16 h at 4°C) to detect the nup82-Δ108-hemagglutinin epitope (HA) fusion protein or with anti-GFP MAb (as above). Immunoblots were developed as described above.

IEM.

For field emission in lens scanning EM (FEISEM), the yeast cell wall was dissolved using Zymolase 20T (Sigma, St. Louis, Mo.) as described previously (51). Yeast nuclei were isolated from spheroplasts by homogenization or by centrifugation of spheroplasts onto a 5- by 5-mm silicon chip for 3 min at 4,000 rpm through buffer A (20 mM Tris-HCl [pH 7.5]), 5 mM MgCl₂, 0.2 M sucrose). Samples were then transferred to buffer A without sucrose, fixed for 15 min in 3.7% formaldehyde, and then given two washes in buffer A without sucrose for 5 min. Samples were blocked for 20 min in 1% BSA–20 mM Tris-HCl, then incubated (with shaking) for 60 min with either the rabbit anti-Nup116-C antibody (1:10 dilution) or the mouse MAb against Nup159p (MAb 165C10; 1:50), and washed twice for 5 min with 0.001% (wt/vol) Tween 20–20 mM Tris-HCl. The samples were subsequently incubated for 30 min with secondary gold-conjugated goat anti-rabbit or anti-mouse antibody (AuroProb TM EM GAR IgG G10; 10 nm in diameter; Amersham) in 0.2% BSA–20 mM Tris-HCl (1:20). The samples were fixed for 10 min in 2% (wt/vol) glutaraldehyde–0.2% (wt/vol) tannic acid in 20 mM Tris-HCl, rinsed with buffer, and incubated for 10 min in 1% OsO₄ in water. Further processing and FEISEM analysis were performed as described previously (18, 38) except that after critical point drying, the samples were coated with 4 nm of chromium. The backscattered electron imaging mode was used to resolve the gold particles by FEISEM, using a solid-state retractable BSE detector in the top stage of the TOPCON (ISI) DS 130F SEM.

For cryo-IEM, yeast nuclei from wild-type cells were purified as described elsewhere (49). Samples containing nuclei in ∼2.5 M sucrose–polyvinylpyrrolidone buffer were diluted with 1 volume of 4% paraformaldehyde–PBS and incubated at room temperature for 2 h. The fixed nuclei were pelleted at 55,000 rpm in a TLA55 rotor for 1 h at 4°C. The pellet was rinsed two times in PBS, embedded in 10% gelatin, and processed for ultracryotomy as described elsewhere (26). Ultrathin sections were prepared and incubated with blocking buffer containing 10% goat serum. Immunolabeling was done with primary antibody (anti-Nup116-C [1:20] or MAb 165C10 [1:5]) for 2 h followed by secondary antibody (12-nm-gold goat-labeled anti-rabbit or anti-mouse IgG) for 1 h. After washing, sections were stained with uranyl acetate and embedded in methyl cellulose (26). Specimens were visualized with a Zeiss-902 EM, and photographs were recorded with Kodak EM film.

RESULTS

A yeast two-hybrid screen with Nup116p-C identifies the nucleoporin Nup82p.

To elucidate the function of the C-terminal region of Nup116p and to identify interacting factors, we conducted a two-hybrid screen using Nup116-C fused to G_BD. A yeast genomic library fused to G_AD was screened to over a 99% confidence level. Positive clones were identified in the two-hybrid host strain, PJ69-4A, which harbors the three Gal4p reporter genes, HIS3, ADE2, and lacZ (36), and tested for specificity by verifying a lack of interaction with G_BD-lamin C. Fourteen library clones were identified, and the genes fused to G_AD were characterized by DNA sequencing. We predicted that such a screen would isolate genes encoding either other nucleoporins or nuclear transport factors. Four of the library isolates contained sequence for NUP82, encoding an essential nucleoporin of 82 kDa (23, 31). To confirm this interaction, the entire coding region of NUP82 was fused to G_AD (pSW1126) and shown to specifically activate HIS3, ADE2, and lacZ by interacting with G_BD–Nup116-C (Fig. 1B, row 1).

FIG. 1.

Two-hybrid interactions between Nup116p and Nup82p. (A) Diagram of the structural regions of Nup116p. N, from residues 1 to 180; GLFG, residues 181 to 725; C, residues 726 to 1113; CN, residues 726 to 919; CC, residues 914 to 1113. (B and C) Mapping the region of Nup116C required for interaction with G_AD-Nup82p and testing other nucleoporins for interaction. Sequences encoding the indicated polypeptides were fused to G_BD and expressed in strain PJ69-4A harboring G_AD-Nup82 or G_AD-Nsp1. Positive interactions in each row are indicated by growth on media lacking histidine and adenine (left) and by the expression of β-galactosidase (right). (D) Diagram of the regions of Nup82p. Amino acid (AA) residues at the deletion points are noted (top, C terminal; bottom, N terminal), and the predicted coiled-coil region spans the C-terminal 200 residues (23, 31). G_AD fusions expressing either C-terminal or N-terminal deletions of Nup82p were constructed and tested for interaction with G_BD–Nup116-C.

Nup116p interacts with the non-coiled-coil region of Nup82p.

To further define the regions of Nup116p and Nup82p that mediate the two-hybrid interaction, a panel of G_AD-Nup82 and G_BD-Nup116 plasmids was constructed and assayed (Fig. 1). The C-terminal half of Nup116-C (Nup116-CC) fused in frame to G_BD interacted with full-length Nup82p to express His3p, Ade2p, and β-galactosidase (Fig. 1B, row 3). In contrast, the N-terminal half of Nup116-C (Nup116-CN) failed to interact with Nup82p (Fig. 1B, row 2). This suggested the C-terminal ∼200 residues were necessary and sufficient for interaction with Nup82p.

The Nup116-CC region harbors a sequence motif that has been reported to bind poly(G) RNA in vitro, designated the NRM (12). To test for the role of the NRM in the Nup116-C–Nup82p two-hybrid interaction, we used three different strategies. First, a G_BD–Nup116-CC fusion was generated wherein the eight amino acids encoding the NRM were replaced with that for a single glycine residue (Nup116-CCΔNRM). The G_BD–Nup116-CCΔNRM fusion did not interact with Nup82p (Fig. 1B, row 4). Second, a G_BD fusion was constructed with sequence encoding only the NRM (G_BD–Nup116-NRM). No interaction was detected between G_BD–Nup116-NRM and G_AD-Nup82 (Fig. 1C, row 2). Third, the Nup100-C and Nup145-M regions also contain NRM motifs, and each show different degrees of structural and functional redundancy with the C-terminal region of Nup116p (12, 63, 66). G_BD–Nup100-C and G_BD–Nup145-M fusions were tested with G_AD-Nup82. The G_BD–Nup100-C fusion interacted with G_AD-Nup82 (Fig. 1B, row 5); however, G_BD–Nup145-M did not (data not shown). Overall, these results suggest that the NRM is not sufficient for interaction and that other sequences conserved only between Nup100-C and Nup116-CC are required.

Previous studies have divided Nup82p into at least two different domains that are both required for its essential function (23, 31), with carboxy-terminal ∼200 residues forming a predicted coiled-coil region. The coiled-coil region is required for isolation of Nup82p with a Nsp1p-Nup159p subcomplex from yeast cells and for the in vitro interaction of Nup82p and Nup159p (6, 23, 32). To determine which region of Nup82p was required for interaction with the C-terminal region of Nup116p, G_AD fusions with deletions from either the N- or the C-terminal end of Nup82p were generated and tested (Fig. 1D). Fusions lacking the predicted coiled-coil region showed an interaction with Nup116-C (Nup82 amino acids 1 to 551 [Nup82:1-551] and Nup82:1-409 [Fig. 1D, rows 2 and 3, respectively]). Further deletions from the C terminus completely abolished the interaction (Nup82:1-264 and Nup82:1-187), and deletions from the N terminus also did not interact (Nup82:482-713 and Nup82:351-713). Thus, the N-terminal Nup82p region comprised of the first 409 residues was necessary and sufficient for interaction with Nup116p. This indicates that the Nup82p structural requirements for interaction with Nup116p are distinct from those needed for Nup82p interaction with Nsp1p and Nup159p.

Given that Nup82p is biochemically isolated with Nup159p and Nsp1p in a subcomplex from yeast cells (6, 23, 32), we tested whether pairwise combinations between different members of this complex could all generate a positive two-hybrid result. A G_AD fusion to Nsp1p and a G_BD fusion to Nup159p were constructed and tested with each other and with the G_BD–Nup116-C and G_AD-Nup82 fusions, respectively. Other combinations have not been analyzed, as expression of full-length G_BD-Nsp1 or G_AD-Nup159 has not been possible. An interaction was observed only between G_AD-Nup82 and G_BD-Nup159 (Fig. 1C, row 1). Finally, others have suggested that Nup82p may interact with the RNA export factor Gle1p (32). However, no positive interaction was observed between G_AD-Nup82 and G_BD-Gle1 (Fig. 1B, row 6).

Isolation of Nup82p and Nup116p in a complex from yeast cells.

Other studies of Nup82p immunoprecipitation complexes have not reported the presence of Nup116p (6, 23). Our previous analysis of protein A-Nup116p complexes documented the coisolation of Kap95p, Gle2p, and a number of other polypeptides (34). Several of the polypeptides are in the 80- to 85-kDa range and could reflect the presence of Nup82p. To directly test whether Nup116p and Nup82p copurify and whether the two-hybrid interaction between Nup116p and Nup82p is physiologically significant, we performed coimmunoprecipitation experiments using whole cell yeast lysates from wild-type cells and from cells expressing Nup82-GFP. Samples were analyzed by immunoblotting with a mouse MAb against GFP (Fig. 2A and B). The Nup82-GFP cells expressed a polypeptide that cross-reacted with the anti-GFP antibody and migrated with a molecular mass of ∼120 kDa (Fig. 2A, lane 2; Fig. 2B, middle, lanes 1, 3, 4, and 7). This band was absent from cells not expressing Nup82-GFP, although the anti-GFP antibody recognized several yeast proteins with lower apparent molecular mass and one larger than 120 kDa (Fig. 2A, lane 1; Fig. 2B, top, lanes 1, 2, 4, 6, and 8).

FIG. 2.

Coimmunoprecipitation of Nup116p and Nup82-GFP from yeast cell lysates. (A) Whole cell lysates from wild-type (W303) (lane 1), Nup82-GFP-expressing (lane 2), or GFP-Nic96-expressing (lane 3) cells were separated by SDS-PAGE and analyzed by immunoblotting with anti-GFP antibody. The Nup82-GFP and GFP-Nic96 fusion proteins migrate at or above, respectively, the 120-kDa marker (indicated at the left). Corresponding bands are not present in the wild-type lysates, although the antibody does recognize several endogenous yeast proteins. (B) Coimmunoprecipitation of GFP-Nup82p and Nup116p. Cell extracts from wild-type, Nup82-GFP, and GFP-Nic96 cells (I, input) were immunoprecipitated with either anti-GLFG, anti-Nup116C, or preimmune serum. Bound (B) and unbound (U) fractions were separated by SDS-PAGE and immunoblotted with anti-GFP antibody. (C) Isolation of Nup82p and Nup116p in a distinct subcomplex. Endogenous yeast proteins in either wild-type (lanes 1 to 4) or NUP82-GFP (lanes 5 to 8) cells were radiolabeled with [³⁵S]methionine and immunoprecipitated with either anti-GLFG, anti-Nup116-C, or preimmune serum, as indicated. The bound fractions were analyzed by SDS-PAGE and autoradiography. Several proteins were present in the anti-GLFG and anti-Nup116-C immunoprecipitations that were not present with the preimmune sera (single arrows, double arrowheads, stars, and circles).

To analyze whether Nup116p interacted with Nup82-GFP, the whole cell lysates from wild-type and NUP82-GFP cells were incubated with affinity-purified rabbit polyclonal antibodies raised against either the GLFG region of Nup116p (Fig. 2B, lanes 2 and 3), Nup116-C (Fig. 2B, lanes 6, 7), or preimmune serum (Fig. 2B, lanes 4, 5, 8, and 9). Coprecipitating proteins were isolated with protein A-Sepharose beads and analyzed by immunoblotting with the anti-GFP antibody to detect Nup82-GFP (Fig. 2B). The bands migrating below 80 kDa or above 120 kDa represent either endogenous yeast proteins recognized by the anti-GFP antibody or IgG. In lysates from wild-type cells (Fig. 2B, top), none of the anti-GFP cross-reactive bands were immunoprecipitated. In contrast, with Nup82-GFP lysates a band at ∼120 kDa representing Nup82-GFP was specifically isolated with both anti-GLFG and anti-Nup116-C antibodies (Fig. 2B, middle, lanes 3 and 7). As a control, Nup82-GFP was not isolated in the bound fraction with the preimmune sera (Fig. 2B, middle, lanes 5 and 9). The same samples were also probed with the anti-Nup116-C antibody to confirm the presence of coimmunoprecipitating Nup116p (data not shown). The immunoprecipitation did not quantitatively isolate either Nup116p or Nup82-GFP from the extracts, as reflected by the presence of Nup82-GFP and Nup116p in the unbound fraction (Fig. 2B, middle, lanes 2 and 6; data not shown). These results suggest that Nup82p and Nup116p can be copurified in a complex from yeast cells.

The anti-GLFG polyclonal antibody recognizes all five members of the GLFG family (8), whereas the anti-Nup116-C antibody is monospecific (33). Therefore, the anti-GLFG immunoprecipitation may coisolate multiple distinct nucleoporin subcomplexes (some which contain Nup116p and some that may not). In addition, based on the two-hybrid results, the Nup82p in the anti-GLFG immunoprecipitations may be associated with Nup100p. However, the anti-Nup116C immunoprecipitation should isolate complexes with only Nup116p. To further analyze the nature of the copurifying complexes, experiments were conducted with a GFP-NIC96 strain. Nic96p is a nucleoporin associated in a complex with Nsp1p and two GLFG nucleoporins, Nup49p and Nup57p (22, 24). Immunoblotting of total yeast cell lysate showed GFP-Nic96 migrating with an apparent molecular mass of slightly >120 kDa (Fig. 2A, lane 3; Fig. 2B, bottom, lane 1). As predicted, GFP-Nic96p was isolated with the anti-GLFG antibody (Fig. 2B, bottom, lane 3). However, GFP-Nic96p was not coprecipitated with the anti-Nup116-C antibody (Fig. 2B, bottom, lane 7). These results suggest that the coisolation of Nup116p and Nup82p with anti-Nup116-C antibody reflects the specific association of Nup116p and Nup82p in a subcomplex in yeast cells.

To more precisely define the composition of the coprecipitating complexes, cell lysates were prepared from both wild-type NUP82 and tagged NUP82-GFP cells after metabolic radiolabeling with [³⁵S]methionine. Coimmunoprecipitations with the anti-GLFG antibody, anti-Nup116-C antibody, and preimmune sera were conducted, and the bound fractions were analyzed by SDS-PAGE and autoradiography (Fig. 2C). For both cell lysates, several polypeptides were specifically isolated in the presence of the anti-GLFG or anti-Nup116-C antibody (Fig. 2C; for wild type, compare lanes 1 and 2 and lanes 3 and 4). Interestingly, a limited number of polypeptides were specifically copurified (Fig. 2C, lanes 1, 3, 5, and 7). Thus, the coimmunoprecipitation strategy is isolating only a subset of nucleoporins. Moreover, with both specific antibodies (Fig. 2C, lanes 1 and 3), polypeptides migrating at ∼120 and ∼82 kDa were present. These likely represent Nup116p and Nup82p, respectively. The conclusion that the band migrating at ∼82 kDa was Nup82p is supported by the results with the Nup82-GFP lysates (Fig. 2C, lanes 5 and 7). The polypeptide at ∼82 kDa was absent in the Nup82-GFP immunoprecipitations, and instead the pattern of bands at ∼120 kDa changed, reflecting an additional polypeptide (Fig. 2C, lanes 5 and 7). This change correlates with the predicted increase in molecular mass for Nup82p with the GFP tag. Finally, there was also at least one distinct difference between the anti-GLFG and anti-Nup116C bound fractions (Fig. 2C, lanes 1 and 5). The anti-GLFG fraction contained a polypeptide migrating at ∼100 kDa, whereas the anti-Nup116-C fraction did not. Based on the analysis of GFP-Nic96 lysates (Fig. 2B, bottom), this polypeptide may represent Nic96p. Taken together, the immunoprecipitation studies strongly suggest that Nup116p and Nup82p can be isolated in a specific subcomplex from yeast cells.

To further confirm the Nup116p-Nup82p interaction, we also tested for coimmunoprecipitation of ³⁵S-labeled Nup82p and Nup116p generated by co-in vitro translation in rabbit reticulocyte lysates. The labeled proteins were coisolated in a complex (data not shown), consistent with the two-hybrid and yeast cell lysate immunoprecipitation results.

Overexpression of Nup116-C in nup116Δ cells results in mislocalization of Nup82-GFP.

The association of Nup82p and Nup116p in a complex suggested that a functional interaction may exist between Nup116p and Nup82p in the NPC. To test if Nup82p localization at the NPC required Nup116p, we examined the localization of Nup82p in cells lacking Nup116p. Nup82-GFP was expressed in a nup116Δ strain, and GFP localization was determined by direct fluorescence of live cells grown at 23°C. In the cells lacking Nup116p, Nup82-GFP was localized at the nuclear rim in a concentrated punctate pattern indicative of NPC localization (Fig. 3A). Therefore, Nup82p incorporation into the NPC does not require Nup116p, although the redundant Nup100p may be sufficient in the absence of Nup116p. Nup82-GFP localization in nup116Δ cells was also examined after shifting the cells to growth at 37°C for 3.5 h. Under these conditions, nup116Δ cells form herniations of the NE over the cytoplasmic face of the NPC (64). After growth at 37°C, the signal for Nup82-GFP was not perturbed and was still observed at the nuclear periphery (Fig. 3B).

FIG. 3.

Nup82-GFP is localized to the NPC in nup116Δ cells. Nup82p fused to GFP was expressed in cells with a NUP116 deletion. The strain was grown at 23°C (top) and visualized for Nup82-GFP localization (left) or shifted to 37°C (bottom) for 3.5 h before visualization at the microscope. Corresponding DAPI staining is shown on the right.

We have previously characterized a nup116-C lethal phenotype wherein expression of the Nup116-CC region in nup116Δ cells results in lethality (28). Expression of the Nup116-CC region in wild-type cells does not affect viability (28). The Nup100-C region also confers lethality when expressed in nup116Δ cells. Interestingly, these are the same regions which interacted with Nup82p (Fig. 1B). This suggested that perturbation of Nup82p (an essential nucleoporin) may be involved in the lethality observed when Nup116-CC is expressed in nup116Δ cells. To analyze the localization of Nup82-GFP in the nup116-C phenotype, GAL-control and GAL-NUP116-C plasmids were transformed into the NUP82-GFP nup116Δ cells. Expression of the Nup116-C was induced by shifting to growth in galactose-containing media. In both glucose and galactose media with the GAL-control plasmid, Nup82-GFP was located at the nuclear periphery in a punctate pattern (Fig. 4A, left). In contrast, shifting to growth in galactose with the GAL-NUP116-C plasmid resulted in a significant decrease in the Nup82-GFP fluorescence intensity and minimal localization at the NE (Fig. 4A, right). Immunoblotting with anti-GFP antibody showed the levels of Nup82-GFP did not change (Fig. 4B). This suggests that expression of Nup116C in nup116Δ cells results in a mislocalization of Nup82-GFP from the NPC and that there is a functional interaction between these two nucleoporins in vivo.

FIG. 4.

Overexpression of Nup116-C in nup116Δ cells results in mislocalization of Nup82-GFP. (A) Localization of Nup82-GFP was analyzed in nup116Δ cells harboring a GAL-control or a GAL-NUP116-C plasmid. All cells were grown to early log phase at 23°C. Growth in glucose (top) resulted in localization of Nup82-GFP at the nuclear rim/NPC. Shifting to growth in galactose for 3 h (bottom) resulted in mislocalization of Nup82-GFP. Minimal staining is observed at the NE. Corresponding DAPI staining is shown on the right. (B) Immunoblot analysis of Nup82-GFP levels. Equal amounts of cell lysates from the indicated samples were separated by SDS-PAGE and immunoblotted with the anti-GFP antibodies. The Nup82-GFP band migrates at ∼120 kDa (arrow). Positions of molecular mass markers are noted in kilodaltons.

Nup116p is specifically mislocalized and degraded in temperature arrested nup82-Δ108 cells.

To determine the localization of Nup116p in the absence of Nup82p, indirect immunofluorescence microscopy experiments were conducted with a temperature-sensitive nup82-Δ108 mutant (31). Localization of Nup116p was detected by indirect immunofluorescence microscopy of nup82-Δ108 cells using the anti-Nup116-C antibody and Texas-red-conjugated goat anti-rabbit antibodies (Fig. 5, left columns). To verify the location of the NPC/NE, the cells were double labeled with mouse MAbs recognizing either Pom152p (61) (Fig. 5A) or Nup159p (39) (Fig. 5B) and FITC-conjugated goat anti-mouse antibodies. In nup82-Δ108 cells grown at 23°C, Nup116p localized at the NE coincident with anti-Pom152p or anti-Nup159p staining (Fig. 5A and B, upper rows). The nup82-Δ108 allele results in a deletion of the C-terminal 108 amino acids of Nup82p (31). Based on our two-hybrid results (Fig. 1D), this region was not required for interaction with Nup116p. Thus, this finding is consistent with the NPC localization of Nup116p in nup82-Δ108 cells at 23°C.

FIG. 5.

NPC localization of Nup116p is perturbed in nup82Δ108 cells, as shown by double-immunofluorescence localization of Nup116p and either Pom152p (A) or Nup159p (B) in nup82-Δ108 cells. nup82-Δ108 cells were grown at 23°C (A and B, top) or shifted to 37°C for 3.5 h (A and B, bottom) and processed for indirect immunofluorescence microscopy. Fixed cells were incubated with affinity-purified rabbit polyclonal anti-Nup116-C and with mouse monoclonal anti-Pom152p (61) (A) or anti-Nup159p (39) (B). The antibodies were detected with Texas red-labeled goat anti-rabbit IgG and FITC-labeled goat anti-mouse IgG. Identical fields are shown in each row, and corresponding DAPI staining is shown on the right.

The nup82-Δ108 strain is temperature sensitive and shows reduced levels of mutant nup82-Δ108 protein after shifting to 37°C (6, 31). To test whether Nup116p was localized at the NPC in the absence of Nup82p, indirect immunofluorescence localization was conducted after the nup82-Δ108 mutant strain was grown at 37°C for 3.5 h. Cells were processed for double labeling with the anti-Nup116-C antibody and anti-Pom152p MAb. At 37°C, the anti-Pom152p staining was at the NE in a punctate pattern typical of NPCs (Fig. 5A, lower panels). In contrast, the anti-Nup116p staining was significantly diminished in a majority of the cells (Fig. 5A, lower panels).

Others have shown that the nucleoporin Nup159p is not localized at the NPC in nup82-Δ108 cells shifted to 37°C (6). To directly compare the behaviors of Nup116p and Nup159p, double staining was conducted (Fig. 5B). As previously reported (6), after growth at 37°C for 3.5 h, the majority of the nup82-Δ108 cells had greatly diminished anti-Nup159p staining at the NE-NPC (Fig. 5B, bottom, middle column). In these same cells, anti-Nup116p staining at the NE-NPC was coincidentally decreased. The phenotype was not completely penetrant, as some cells showed weak anti-Nup159p and/or anti-Nup116p staining at the nuclear rim (Fig. 5B, bottom). Overall, the anti-Nup159p staining appeared to be more readily mislocalized than the anti-Nup116p staining. Given that both Nup116p and Nup159p levels were diminished, it was possible that the nup82-Δ108 phenotype resulted in indirect perturbations on all peripheral nucleoporins. Nic96p is isolated in a subcomplex containing Nsp1p (22, 24). As a control, the localization of GFP-Nic96 in nup82-Δ108 cells was analyzed. After growth at 37°C, the signal for GFP-Nic96 remained at the NPC and was not perturbed (Fig. 6). The specific localization perturbations observed in nup82-Δ108 cells suggest that Nup82p plays a direct role in the NPC association of both Nup116p and Nup159p.

FIG. 6.

GFP-Nic96 localization at the NE-NPC is not perturbed in nup82-Δ108 cells at 37°C. SWY2126 cells were grown at 23°C (top) and then shifted to 37°C for 3.5 h (bottom). GFP-Nic96 localization was visualized by direct fluorescence microscopy (left); the corresponding Nomarski field is shown on the right.

To monitor the protein levels of Nup116p in nup82-Δ108 cells at 37°C, immunoblotting experiments were conducted (Fig. 7). As previously reported, the majority of the mutant nup82-Δ108 protein was degraded after ∼3 h at 37°C (6, 31). In a parallel manner, Nup116p levels also decreased. Interestingly, all NPC-associated proteins were not perturbed by degradation of nup82-Δ108 protein. The levels of Nup57p appear stable over the same time course (Fig. 7, bottom).

FIG. 7.

Nup116p is degraded in nup82-Δ108 cells at 37°C. Cells were grown at 23°C to an OD₆₀₀ of ∼0.5 and then shifted to 37°C for the indicated times; 25 mg of each sample was lysed, and equivalent fractions were separated by SDS-PAGE. Immunoblots were conducted with MAb 12CA5 to detect the Nup82-Δ108–HA fusion protein (top), anti-Nup116-C antibody (middle), or anti-GLFG antibody (bottom). Positions of molecular mass markers are noted in kilodaltons.

Nup116p is localized on both the cytoplasmic and nuclear faces of the NPC.

Previous IEM studies by others have shown that Nup82p is localized exclusively on the cytoplasmic side of the NPC (13, 32). Although IEM experiments with epitope-tagged Nup116p have been conducted (64), the morphological resolution of the previous studies was not sufficient to determine the substructural localization of Nup116p. To localize Nup116p, we used two different IEM strategies with wild-type yeast cells and the anti-Nup116C antibody. First, we combined FEISEM with immunogold labeling. FEISEM allows samples of infinite depth to be surface imaged, and striking images of vertebrate NPCs have been obtained (17, 19). Recently similar methods have been applied to yeast nuclei (E. Kiseleva et al., unpublished data), showing the NE cytoplasmic surface covered with ribosomes and the NPCs as small invaginations surrounded by short filaments. To localize Nup116p using FEISEM, nuclei from wild-type and nup116Δ cells were isolated, fixed, and incubated with the rabbit Nup116-C antibody, followed by 10-nm gold-labeled anti-rabbit antibodies. In Fig. 8, two panels are shown for each field: the cytoplasmic face of the nuclear envelope and NPCs observed by FEISEM (Fig. 8a and c), and gold labeling revealed by FEISEM with backscattered electron imaging (Fig. 8b and d). Anti-Nup159p labeling was used as a control to confirm the structures represent NPCs (data not shown). In wild-type cells, gold labeling was present at sites concident with NPC-like structures (Fig. 8a and b). In contrast, no specific gold labeling was observed at NPC-like structures in nuclei from nup116Δ cells (Fig. 8c and d) or after incubation of wild-type cells with secondary antibodies alone (data not shown). Thus, Nup116p is localized to at least the cytoplasmic face of the NPC.

FIG. 8.

Nup116p localizes to the cytoplasmic face of the NPC in the FEISEM experiments. Nuclei from wild-type (a and b) or nup116Δ (c and d) cells were isolated, fixed in 3.7% formaldehyde, incubated with anti-Nup116-C antibody and 10-nm gold-conjugated anti-rabbit secondary antibodies, and processed for FEISEM. Images show the morphology of the cytoplasmic face of the nucleus (a and c) and the location of the gold particles in the same field (b and d). Gold particles are indicated by arrows in the panel b, and the corresponding location on the NPC is indicated by arrows in panel a. Arrowheads indicate the position of NPCs without gold labeling in nup116Δ cells (c). Bars = 66.7 (a and b) and 125 (c and d) nm.

The second approach used cryo-IEM with purified wild-type yeast nuclei. The sections were incubated with the anti-Nup116-C antibody (Fig. 9) or the anti-Nup159p MAb (data not shown) and the appropriate secondary antibodies coupled to 12-nm gold particles. Labeling was not observed when the samples were incubated with only the secondary antibodies (data not shown). The number of gold particles and their distance from the midplane of the nuclear pore membranes were scored in multiple sections, with particles noted as being either at the midplane (+ or −25 nm) or on the cytoplasmic or nuclear side (more than 25 nm from the midplane). For anti-Nup116-C labeling (n = 135 particles), the gold particles were present at the NPCs on both sides of the NE. However, the distribution was asymmetric, with the majority of the labeling (54%) on the cytoplasmic face and the remainder split between that at the midplane itself (18%) or on the nuclear face (27%). In contrast, the anti-Nup159p labeling was localized almost exclusively on the cytoplasmic face of the NPCs (n = 104; 83% on the cytoplasmic face, 12% at the midplane, and 5% on the nuclear face). This is consistent with previous reports of Nup159p localization (39). While this work was under review, others reported a similar IEM localization pattern for protein A-tagged Nup116p on isolated NEs (48). Thus, Nup116p is localized on both the cytoplasmic and nuclear faces of the NPC, with a majority positioned on the cytoplasmic face.

FIG. 9.

Cryo-IEM localization of Nup116p at both nuclear (n) and cytoplasmic (c) faces of the NPC. Purified wild-type nuclei were fixed and processed for cryo-IEM with anti-Nup116-C antibody and 12-nm gold-conjugated anti-rabbit antibodies. Representative micrographs of NE spans are shown with arrowheads at the gold particles (multiple gold particles often label a single NPC; open arrow in panel A). Bar = 100 nm.

DISCUSSION

Understanding the nearest-neighbor protein-protein interactions among components of the NPC will be critical for defining the mechanism of nuclear transport. Nup116p plays a central role in both nuclear import and export. Shuttling nuclear transport factors have been shown to interact with both the N-terminal and GLFG domains of Nup116p (1, 4, 28, 33, 50, 56). Here, we show that the C-terminal region of Nup116p is associated with Nup82p. Nup82p was identified in a two-hybrid screen with Nup116C as a bait. These two nucleoporins specifically copurified by immunoprecipitation from whole yeast cell lysates. Moreover, each was mislocalized from the NPC in particular mutant backgrounds. The nup116C lethal phenotype results in mislocalization of Nup82-GFP, and degradation of nup82-Δ108 results in the specific mislocalization and coincident instability of Nup116p. We also present evidence that endogenous Nup116p is localized on both faces of the NPC, with a majority at the cytoplasmic face. Taking these findings together, we propose Nup82p associates with several nucleoporins at the cytoplasmic face and as such positions these nucleoporins for executing key steps in mRNA export.

Others have reported that the C-terminal region of Nup116p is required for targeting Nup116p to the NPC (4). Such a docking function correlates with the requirement for the Nup116p C-terminal region in normal yeast cell growth (33). However, the mechanism for Nup116p assembly into the NPC and the nearest-neighbor NPC proteins for Nup116p had not been previously characterized. There are significant mechanistic implications for the role of Nup116p in nuclear transport based on its substructural location in the NPC, and the interaction between Nup116p and Nup82p establishes a further link between a subset of NPC components and nuclear transport factors. Nup82p is localized exclusively at the cytoplasmic face of the NPC (32), and the temperature-sensitive nup82-Δ108 mutant has defects in mRNA export (31). Nup159p is exclusively localized to the cytoplasmic NPC face (39), and it is specifically required for mRNA export (20). The associations within the Nup82p-Nsp1p-Nup159p complex are likely mediated through coiled-coil interactions of heptad repeat domains within each protein (6, 23, 32). Interestingly, the interaction of Nup116p with Nup82p requires only the N-terminal non-coiled-coil region of Nup82p. Therefore, associations of Nup82p with Nup116p and Nup159p are not mutually exclusive.

Interaction of Nup116p with Nup82p would place the N-terminal and GLFG regions of Nup116p as cytoplasmic docking sites for both the mRNA export factor Gle2p and members of the karyopherin family, respectively. Recent studies have shown that Nup159p serves to recruit the RNA helicase Dbp5p to the NPC (29, 55). Dbp5p shuttles between the nucleus and cytoplasm and is required for mRNA export (29, 55, 59, 62). It is exciting to speculate that coincident Nup159p and Nup116p docking at Nup82p would effectively juxtaposition binding sites for several RNA export factors at one NPC substructure. This positioning, constrained by the exclusively cytoplasmic localization of Nup82p, may explain why the phenotypes of nup116, nup82, and nup159 mutants are all linked to mRNA export. The coordinated action of Dbp5p docked at Nup159p and Gle2p at Nup116p may be required for a distinct mRNA export step at the cytoplasmic face.

Although the Nup82p-Nup116p interaction is specific, our results also indicate that Nup116p may have additional NPC binding partners. First, the IEM analysis showed that Nup116p is also localized on the nuclear face of the NPC; however, Nup82p is exclusively cytoplasmic (32). Second, in the nup82-Δ108 mutant at the restrictive temperature, the localization of Nup159p at the NPC is more readily perturbed than Nup116p. Third, in a nup57 temperature-sensitive mutant, Nup116p is mislocalized to the cytoplasm even though Nup82p is not perturbed (8). Interestingly, the stability of Nup57p is not altered in the nup82-Δ108 mutant. Even though Nup82p is not present on the nuclear side, it may affect Nup116p at the nuclear face. Although a significant amount of Nup116p is localized on the nuclear face (Fig. 9), the total Nup116p pool degrades at the same rate as the nup82-Δ108 pool at 37°C (Fig. 7). It is possible that Nup116p is dynamic within the context of the NPC structure, as has been recently suggested for the vertebrate GLFG nucleoporin Nup98p (70), and the cytoplasmic Nup82p could then interact with the total Nup116p pool. Alternatively, Nsp1p may provide a common link between the nup57 and nup82 mutant phenotypes and their perturbations of Nup116p. Nsp1p is symmetrically localized on both sides of the NPC (48), and it can be isolated in a complex with Nup82p and Nup159p (6, 23). Nsp1p also copurifies with Nup49p, Nup57p, and Nic96p in a stable complex (22, 24, 54). The mislocalization of Nup116p in the nup57 mutant cells is coincident with a perturbation of Nsp1p (8), and nup116Δ cells are synthetically lethal with nsp1 mutants (67). This suggests that Nup116p and Nsp1p are functionally, if not physically, associated. However, Nsp1p and Nup116-C do not interact in the two-hybrid assay (Fig. 1C). Defining additional protein-protein interactions for Nup116p at the NPC will be a focus of future studies.

Based on the interaction observed between Nup100p and Nup82p in the two-hybrid assay, Nup100p may also be incorporated into the NPC by interactions with Nup82p. In combination with the extensive genetic connections between NUP100 and NUP116 (4, 12, 28, 34, 63), these results further suggest that Nup100p could be asymmetrically localized on both the cytoplasmic and nuclear faces similarly to Nup116p. This prediction has been confirmed in a recent publication (48). In contrast, Nup145-M appears to be functionally distinct and localized differently from Nup100p and Nup116p (48). This is also suggested from the lack of interaction between the G_BD–Nup145-M and G_AD-Nup82. Moreover, Nup100-C and Nup116-C are more closely related to each other by protein sequence homology than either is to Nup145-M (63). Due to the caveats associated with the two-hybrid assay, the apparent lack of interaction between Nup145-M and Nup82p will need to be investigated further.

A previous study has proposed that the NRM motif in the C-terminal region of Nup116p directly binds RNA during nuclear transport (12). This is based on in vitro binding of Nup116-C and Nup145-M to poly(G) in vitro (12). The Nup116-C, Nup100-C, and Nup145-M regions probably participate in multiple functions at the NPC. However, the physiological significance of the poly(G) binding activity for Nup116p, Nup100p, and Nup145p has not been demonstrated. It is unclear how the NRM-containing region participates in binding both RNA and Nup82p, and it is not known whether Nup116p can interact with both simultaneously. It is intriguing to consider that Nup116p may interact with RNA on the nuclear face and with Nup82p on the cytoplasmic face of the NPC.

We had previously assumed that Nup116p was localized exclusively on the nuclear face of the NPC (34). This hypothesis was based on the localization of the single vertebrate GLFG nucleoporin Nup98p exclusively at the nuclear NPC face (45) and on the fact that Nup98p and Nup116p appear highly similar in terms of sequence, Gle2p interaction, and karyopherin binding (4, 15, 28, 33, 43–45). The recent study suggesting Nup98p is dynamic may allow Nup98p to also associate with the cytoplasmic NPC face (70). Knowing Nup116p is on both faces of the NPC presents a change in thinking about its role in the nuclear transport mechanism. Our current focus involves further delineation of nearest-neighbor interactions for each component within the NPC and elucidating how the structure facilitates movement through the NPC.